CCVisu: A Tool for Visual Graph Clustering and General Force-Directed Graph Layout

CCVisu: A Tool for Visual Graph Clustering
and General Force-Directed Graph Layout

Software Systems Lab
Univerisity of Passau, Bavaria, Germany

Objectives

General force-directed graph layout.

The tool CCVisu is a light-weight tool for force-directed graph layout. The tool reads the input graph from a file in RSF (Relational Standard Format), which is a standard text format for relations. The layout of the graph is computed using standard techniques from force-directed layout. The tool supports several energy models, which can be selected by setting command line parameters. The weighted edge-repulsion LinLog energy model (default) is good for producing layouts that fulfill certain clustering criteria. The Fruchterman Reingold energy modelis good for producing layouts that fulfill certain esthetic criteria like uniform edge length. The minimizer is based on the Barnes-Hut-Algorithm. CCVisu saves the resulting layout in the file formats SVG, VRML, the standard text format for relations RSF, or it displays the layout on the screen.
Software-Structure Extraction.

Structure analysis requires a preprocessing tool, called fact-extractor, which extracts from a software system the structure, i.e., abstract dependency graphs such as call graphs, inheritance graphs, and containment graphs. CCVisu implements two kinds of fact-extractors: the co-change graph extractor for version history and the Doxygen extractor for structure and dependency information (cf. CCVisu input formats).
Software Decompostition.

Clustering layouts of software systems combine two important aspects: they reveal groups of related artifacts of the software system, and they produce a visualization of the results that is easy to understand. The energy model that produces clustering layouts decomposes the software graph into cohesive subsystems, and separates loosly coupled parts of the system. The tool is designed as a framework, easy to use, and easy to integrate into reengineering environments; several formats for data interchange are already implemented.
Co-Change Visualization.

Co-change visualization is a lightweight method for computing clustering layouts of software systems for which the change history is available. The tool implementation CCVisu extracts the co-change graph from a CVS or SVN version repository, and computes a clustering layout based on energy models, which positions the artifacts of the software system in a two- or three-dimensional space. Two artifacts are positioned closed together in the layout if they were often changed together.

CCVisu is free software (LGPL).

New: Project page available at Google Code: http://ccvisu.googlecode.com/
(includes source code repository, downloads, documentation)

CCVisu Introduction and Reference Manual (full document as PDF or HTML, table of contents, example files)

Getting started (short tutorial)

Download Java sources, jar file, and example data

FSF Free Software Directory: http://directory.fsf.org/ccvisu.html

Publications

Full references and electronic versions of the papers are available at our Publications page.

Dirk Beyer. CCVisu: Automatic Visual Software Decomposition. ICSE 2008.
Tool paper for the software implementation CCVisu, as a tool for general force-directed graph layout.
Dirk Beyer. Co-Change Visualization Applied to PostgreSQL and ArgoUML. ICSE-MSR 2006.
Application of co-change visualization to the software systems PostgreSQL and ArgoUML.
Supplementary material.
Dirk Beyer. Co-Change Visualization. ICSE 2008.
Describes a software implementation of the IWPC'05 paper that can be used for Co-Change Visualization.
Dirk Beyer, Andreas Noack. Clustering Software Artifacts Based on Frequent Common Changes. IWPC 2005.
Introduces the method, consisting of the co-change graph and the edge-repulsion LinLog energy model. Discusses three examples.
Supplementary material.

Author: Dirk Beyer, my project pages: BLAST, CCVisu, Chic, CrocoPat, CSIsat, Rabbit